H-1B Visa Employers

Circle packing

Photo by Kate Sade on Unsplash

…in our increasingly global economy, highly skilled foreign workers are certain to be in a position to make unique contributions to the U.S. economy…
— Senate Judiciary Committee 2000

The H-1B is a visa in the United States under the Immigration and Nationality Act, section 101(a)(15)(H) that allows U.S. employers to temporarily employ foreign workers in specialty occupations. A specialty occupation requires the application of specialized knowledge and a bachelor’s degree or the equivalent of work experience. The duration of stay is three years, extendable to six years; after which the visa holder may need to reapply. Laws limit the number of H-1B visas that are issued each year: 188,100 new and initial H-1B visas were issued in 2019.

To illustrate the circle packing layout and custom configurations to manage the large number of data points, we’ll demonstrate these key features:

Create unique node ids to collapse the hierarchy, but retain the values
Apply conditional label formatting (what you see as a label)
Apply conditional label visibility (when labels are shown)

Ingest the data

employer, state, and approvals

df_file_path <- "archetypes/h-1b-employers/2019-H-1B-Employers.csv"
df = read.csv(df_file_path, header = TRUE, stringsAsFactors = FALSE)
df

Wrangle the data

Create node and edge tables

# Complete cases
df_wrangle <- df %>% mutate(ID = row_number(), Approvals = as.integer(Initial.Approvals)+as.integer(Continuing.Approvals) )
df_wrangle <- filter(df_wrangle, nchar(State) > 0 )
df_wrangle <- filter(df_wrangle, nchar(Employer) > 0 )
df_wrangle <- filter(df_wrangle, nchar(ZIP) > 0 )

# limited for performance
df_wrangle <- filter(df_wrangle, Approvals > 2 )

# create a unique node id to collapse three level hierarchy
df_wrangle <- df_wrangle %>% mutate(EMP_UNIQ = paste0(Employer, "_", State, "_", ZIP))

# unique edges
df_edges <- aggregate(x = df_wrangle$Approvals,
          by = list(df_wrangle$State, df_wrangle$EMP_UNIQ),
          FUN = sum)

# standard edge table structure
colnames(df_edges) <- c("FROM","TO", "SIZE")
# df_edges

# root nodes
df_nodes_1 <- aggregate(x = df_wrangle$Approvals,
          by = list(df_wrangle$State),
          FUN = sum)

colnames(df_nodes_1) <- c("NODE","SIZE")
# df_nodes_1

# leaf nodes
# df_nodes_2 <- df_wrangle %>% select(EMP_UNIQ, Approvals)
df_nodes_2 <- aggregate(x = df_wrangle$Approvals,
         by = list(df_wrangle$EMP_UNIQ),
          FUN = sum)
colnames(df_nodes_2) <- c("NODE", "SIZE")
# df_nodes_2

# Combine
df_nodes <- rbind(df_nodes_1, df_nodes_2)

# Transform to graph data structure
df_graph <- graph_from_data_frame( df_edges, vertices = df_nodes )

edge table

df_edges

node table

df_nodes

Plot

circle pack layout with size, color, and labels (at root level)

theme_opts <- theme(
    text = element_text(family = "inconsolata"), 
    legend.position='none'
  )

n <- 15
top_x <- head(arrange(df_nodes_2, desc(SIZE)), n = n)
top <- top_x$SIZE[[1]]
bottom <- top_x$SIZE[[n]]

v1 <- ggraph(df_graph, layout = 'circlepack', weight = SIZE) + 
  geom_node_circle(fill="#F0F0F0") + 
  geom_node_label( aes(label=name, filter=depth==0), size = 6, family = "inconsolata") +
  geom_node_text( aes(label=gsub("_", "\n", name), filter=SIZE >= bottom & SIZE <= top), size = 3, color = "#333333", family = "inconsolata", face="bold") +
  coord_fixed() + 
  theme_void() +
  theme_opts

girafe(ggobj = v1, width_svg = 1280/72, height_svg = 1280/72,
       options = list(opts_sizing(rescale = TRUE, width = 1.0))
)

References

citations for narrative and data sources

Narrative sources: Mike Bostock, Observable, https://observablehq.com/@mbostock/2019-h-1b-employers
Data sources: USCIS, H-1B Employer Data Hub, https://www.uscis.gov/tools/reports-and-studies/h-1b-employer-data-hub